Statistical analysis
This analysis is essentially the same as the first, where I ran both Bayesian and Frequentist mixed effects logistic ordinal regression models. Again, the ordered factor rating (1-5) was the outcome variable, and sentence type was the predictor. In this case, all 4 languages had the same sentence types, in which ad adverb was pre or post verb in a either a main or relative clause. Just like the first dataset, random intercepts were added per participant in order to account for the nested structure of the data in both types of model.
Results
The tables below detail the descriptive statistics for each sentence type and the quantity of ratings for each condition.
| rating | main_post_verb | main_pre_verb | relative_post_verb | relative_pre_verb |
|---|---|---|---|---|
| 1 | 5 | 3 | 3 | 1 |
| 2 | 7 | 1 | 3 | NA |
| 3 | 7 | 5 | 3 | 4 |
| 4 | 4 | 6 | 8 | 11 |
| 5 | 3 | 11 | 9 | 9 |
| rating | relative_pre_verb | relative_post_verb | main_pre_verb | main_post_verb |
|---|---|---|---|---|
| 1 | 2 | 3 | 3 | 8 |
| 2 | 1 | 5 | 2 | 10 |
| 3 | 4 | 1 | 5 | 5 |
| 4 | 9 | 11 | 8 | 2 |
| 5 | 11 | 7 | 9 | 2 |
| rating | relative_pre_verb | relative_post_verb | main_pre_verb | main_post_verb |
|---|---|---|---|---|
| 1 | 12 | 18 | 11 | 6 |
| 2 | 9 | 16 | 18 | 8 |
| 3 | 8 | 11 | 11 | 7 |
| 4 | 18 | 14 | 19 | 23 |
| 5 | 28 | 16 | 16 | 31 |
Visualization of results
Same as the first analyiss, the two plots below show the output of a Bayesian Ordinal Logistic regression (in the first plot), and probability curves per condition in the second plot. In plot 1, the points represent the probability (y-axis) and 95% confidence intervals per condition (x-axis). In the second plot, the vertical lines represent each condition, where the individual lines are probabilities of a rating per condition. The closer the vertical lines are together, the closer those conditions are rated in a particular language.
Here are the file names of the plots in this page:
Polish plot 1 = polish_word_order.png
Polish plot 2 = polish_prob_curve_word_order.png
Norwegian plot 1 = norwegian_word_order.png
Norwegian plot 2 = norwegian_prob_curve_word_order.png
English plot 1 = english_word_order.png
English plot 2 = english_prob_curve_word_order.png
Posterior distributions of Polish ratings per conditions based on Bayesian Ordinal Regression 
Posterior distributions of Norwegian ratings per conditions based on Bayesian Ordinal Regression 
Posterior distributions of English ratings per conditions based on Bayesian Ordinal Regression 
Visualizations per condition between languages
As we discussed, here are the ratings of each condition per language. Again, probability is on the y-axis and each color represents a rating. The error bars are the 95% confidence interval of the estimates produced by the Bayesian ordinal regression model. You could save these by right clicking, but the files are also available in the repository under plots/word_order
File names:
main_pre.png
main_post.png
rel_pre.png
rel_post.png
